You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In case no other source selection mechanism is used (no alpha, no ROI, no NoData) the mosaic will simply generate in output its first input image, with padding.
This is how it works:
Each input image is padded to the extends of the output via a BorderOpImage (inefficient, the original operation used source.getExtendedData instead, which could use optimizations such as generating child rasters or use arraycopy to copy over data, while the BorderOpImage does a pixel by pixel check (for nodata) and copy
While looping over the data, if no other mechanism to detect input validity is used, then the pixel is unconditionally accepted as valid (the original mosaic op used the thresholds instead), even if it was not part of the original image
The thresholds are not used at all (see #224) but regardless of that, the code should be checking the input raster extents and avoid pulling pixels out of extended pixels, shouldn't it?
The text was updated successfully, but these errors were encountered:
Going to make it work over the original source rectangle, skipping pixel processing for sources that are not providing any actual value for the current position
aaime
added a commit
to aaime/jai-ext
that referenced
this issue
Dec 5, 2018
In case no other source selection mechanism is used (no alpha, no ROI, no NoData) the mosaic will simply generate in output its first input image, with padding.
This is how it works:
The thresholds are not used at all (see #224) but regardless of that, the code should be checking the input raster extents and avoid pulling pixels out of extended pixels, shouldn't it?
The text was updated successfully, but these errors were encountered: